Platform · Integrations

Drops into the stack you already run. No agent rebuilt.

TapPass governs the models and MCP servers your agents already use, streams the record into the SIEM you already run, and signs people in through your own identity provider.

Start free Talk to our team

✓One base URL · any OpenAI-compatible SDK

16model providers and routers, through one base URL

13coding agents, IDEs and agent frameworks

AnyMCP server, with permissions per tool

OTLPevery decision, streamed to the SIEM you run

01The directory

Works with what you run. Filter by where it plugs in.

If it speaks the OpenAI API or MCP, it works. For anything else, call the Verdict API before the agent acts.

Models and routersAny OpenAI-compatible model through one base URL. Keep the router you run.

OpenAIGPT models through the gateway

AnthropicClaude models through the gateway

MistralEU-hosted models for personal data

Google GeminiGemini models through the gateway

Azure OpenAIYour own Azure deployments

AWS BedrockModels in your AWS account

Meta LlamaOpen-weight models

DeepSeekThrough the gateway, by your rules

xAI GrokThrough the gateway, by your rules

KimiThrough the gateway, by your rules

PerplexitySearch models through the gateway

Hugging FaceHosted and self-hosted models

OllamaModels on your own servers

OpenRouterKeep your router behind TapPass

LiteLLMKeep your router behind TapPass

PortkeyKeep your router behind TapPass

Agents, IDEs and frameworksCoding agents and agent frameworks, governed without a rebuild.

Claude CodeEvery prompt and tool call, through its hooks

OpenAI CodexThrough the gateway and MCP

GitHub CopilotMCP tools governed per agent

CursorMCP tools governed per agent

VS CodeMCP tools governed per agent

WindsurfMCP tools governed per agent

JetBrainsMCP tools governed per agent

ReplitMCP tools governed per agent

OpenAI Agents SDKOne base URL, no code change

LangChainEvery chain and tool call routes through

CrewAISame base URL, same rulebook

LlamaIndexSame base URL, same rulebook

Vercel AI SDKStreaming included

MCP serversPer-tool permissions on any MCP server, or a virtual server with only the tools a job needs.

GitHubRepositories, issues and pull requests

LinearIssues and projects

JiraIssues and projects

NotionPages and databases

FigmaDesign files

ObsidianNotes and vaults

Any MCP serverTools classified by intent, then granted

Business systemsWhere agents act on real work. Payments, messages and exports can wait for a person.

SlackMessages, and approval notifications

Microsoft TeamsApproval notifications

Microsoft 365Mail, files and calendars

GmailMail, and approval notifications by email

Google DriveFiles and documents

StripePayments and refunds, held above your limit

MailchimpAudiences and campaigns

SIEM and observabilityEvery decision streams out as OpenTelemetry, with sensitive content redacted on the way.

SplunkDecisions as OpenTelemetry

Microsoft SentinelDecisions as OpenTelemetry

ElasticDecisions as OpenTelemetry

OpenTelemetryAny OTLP sink you already run

API gatewaysKeep your API gateway. Call the Verdict API from it, or put TapPass behind it.

KongAsk the Verdict API before forwarding

MuleSoftAsk the Verdict API before forwarding

Azure API ManagementAsk the Verdict API before forwarding

Identity and sign-inPeople sign in through your identity provider and approve with a passkey on their own device.

OktaSingle sign-on

Microsoft Entra IDSingle sign-on

Google WorkspaceSingle sign-on

Any SAML 2.0Single sign-on

PasskeysWebAuthn approvals on the approver’s own device

itsmesoonStrong-auth approvals

02Two ways to connect

A proxy, or a question from your code. Same rules, same record.

.env

export OPENAI_BASE_URL="https://<your-tappass>/v1"
export OPENAI_API_KEY="<your TapPass key>"

Gateway

Repoint one base URL. Every model and MCP tool call passes your rules before it runs. No application changes.

Explore the gateway →

POST /v1/govern

{ "type": "TOOL_CALL",
  "agent_id": "procurement-agent",
  "payload": { "tool": "kbc.create_payment" } }

Verdict API

Ask before an agent acts and enforce the answer yourself. The fit for traffic that stays behind Kong, MuleSoft or your own code.

Explore the Verdict API →

See how the pieces fitOne rulebook and one signed record across the gateway, MCP gateway and Verdict API.

Questions

Integrations, in plain answers.

Something else on your mind? Talk to our team.

Do we have to change our agents?

No. Point the base URL of any OpenAI-compatible SDK or framework at TapPass and use your TapPass key. MCP clients connect to TapPass instead of to each server.

Our tool is not on this page. Will it work?

If it speaks the OpenAI API or MCP, yes. For anything else, call the Verdict API before the agent acts. Ask us if you are unsure.

Can we keep our own model router?

Yes. Keep LiteLLM, OpenRouter, Portkey or your own router. TapPass sits in front of it and decides before the call goes on.

How does the record reach our SIEM?

Every decision streams out as OpenTelemetry to Splunk, Microsoft Sentinel, Elastic or any OTLP sink. Sensitive fields are removed before anything leaves TapPass.

Keep your stack. Add the rules it is missing.

Connect one agent and see every call TapPass would have held, stopped or routed elsewhere. Nothing is blocked while you learn.

Start free Talk to our team

✓ Free for one person✓ No credit cardBacked by

✓ EU hosted✓ Built in Belgium✓ Your own model keys✓ Encrypted at rest and in transit✓ DPA on request